response such as pitch, structural stresses, etc., 

 the distribution of peaks may be expected to de- 

 pend upon the characteristics of the ship's fre- 

 quency-response functions assuming that the ship 

 is a simple linear system. As long as the ship re- 

 sponse involves a single degree of freedom that is 

 relatively lightly damped (and thus yielding a 

 narrow and peaked frequency-response function), 

 the Rayleigh distribution may still be expected 

 to apply. However, if the ship response is more 

 complicated involving several significant degrees 

 of freedom, the resulting power spectrum of the re- 

 sponse may no longer contain a simple peak but 

 may contain multiple peaks. For such cases, the 

 Rayleigh distribution would not appear to be ade- 

 quate, and the more general distribution given by 

 Rice would have to be used. 



The writer is not familiar with how serious the 

 foregoing objections actually may be in practical 

 applications. The experimental data presented 

 do, however, show sufficient departure from the 

 assumed hypotheses to warrant a more detailed 

 consideration of the foregoing types of limitations. 



In regard to the second major point, the ap- 

 plicability of the log-normal distribution, we are 

 here concerned with a distribution of peaks for the 

 summation of a large number of stationary Gaus- 

 sian processes. There does not appear to be any 

 theoretical a priori basis for the choice of a distri- 

 bution function, and one is forced here to rely more 

 heavily on the purely empirical observations. 



The empirical data show generally good ad- 

 herence to the assumed log-normal distribution 

 function, but again the data appear to show de- 

 partures from log-normal distributions that sug- 

 gest the possibility of the need for a more detailed 

 consideration. It appears clear that the appro- 

 priate distribution may well be expected to vary 

 depending upon the detailed nature of the opera- 

 tions being considered and the climatology of the 

 ocean waves. In an analogous problem in airplane 

 studies, no single distribution was found to apply 

 in general to gust velocities encountered by air- 

 planes and to airplane response distributions under 

 full operating conditions. 



One further point is worth noting. The author 

 has chosen to describe the ocean characteristics in 

 terms of the probability distribution of the ocean 

 wave heights. Although this is a useful descrip- 

 tion for many purposes, it is not adequate in re- 

 gard to the problem of determining the ship re- 

 sponse since the response of a ship will depend 

 upon not only the individual wave heights but 

 also upon the preceding waves or more concisely 

 upon the power spectrum of the ocean wave 

 height. Thus, a more fundamental and useful de- 

 scription of disturbed seas would appear neces- 



sary in the long run if adequate predictions of ship 

 behavior in operations are to be achieved. 



In concluding, I should like to reiterate the 

 valuable nature of the present paper. It cer- 

 tainly provides some reasonable and useful defini- 

 tions of the ocean wave field and the character of 

 the ship response. The principal criticisms in the 

 foregoing are concerned with the absence of ex- 

 pressed limitations to the hypotheses claimed and 

 the suggestion is that refinements to the simple 

 picture presented may often become necessary. 



CoMDR. R. L. Brooks, UvSN, Associate Mem- 

 ber: This paper provides a working tool of great 

 value to the naval architect for obtaining, in- 

 terpreting, and using full-scale data. By means of 

 the distribution pattern and with a minimum of 

 observations, many seaworthiness problems can 

 be clarified in terms of actual results obtained at 

 sea on a properly chosen hull. 



That the need for reliable data respecting ship 

 responses to the sea is a matter of great impor- 

 tance has become better appreciated, as the solu- 

 tion of each new seaworthiness problem was found 

 to depend on full-scale data which were seldom 

 readily available in the form or ranges which were 

 needed. Examples of critical problem areas in 

 which progress no longer needs to be hampered by 

 the lack of full-scale data, now that this distribu- 

 tion pattern technique is available, include: (a) 

 The determination of how much ships must slow in 

 heavy seas, ib) the checking of assumptions upon 

 which are based theories for predicting ship 

 motions, (c) the improving of present strength 

 design methods, and {d) the development of ship- 

 board computers for selecting optimum courses 

 and speeds for a variety of tactical situations, 

 as for example fueling of destroyers in moderate 

 to heavy seas. 



While the need for rehable data has long been 

 acknowledged, the difficulties which stood in the 

 way of obtaining it at sea, have been so serious 

 as to hold up solutions to many of the important 

 problems for several years. 



Consideration of a few of these difficulties helps 

 to show that the present paper represents a task 

 of large magnitude, because each difficulty was a 

 major problem in its own right. The following 

 examples are mentioned: (a) The development 

 of specialized recording instruments which would 

 require a minimum of attention, {b) the planning 

 of how the many variables which define the 

 environment and the ships' responses should be 

 handled, (c) the detailed co-ordination of in- 

 strument installation and operation, so as to 

 harmonize with ship schedules and internal 

 routine, (d) the difficulty of defining each sea 



52 



